If you’re involved in electronics, you’ll probably have an oscilloscope on your bench. As electronics become more complex almost daily, sooner or later a new oscilloscope will be in order. How to choose the right one for your applications?
Factors to consider:
1 Remember that the bandwidth specification of an oscilloscope is the frequency of the “-3 dB point” of a sine-wave signal of a particular amplitude, e.g. 1 Vpp. As the frequency of your sinewave goes up (while keeping the amplitude constant), the measured amplitude goes down. The frequency at which this amplitude is -3 dB lower, is the instrument’s bandwidth. This means that an oscilloscope of 100MHz would measure a 1Vpp sinewave of 100MHz at only (approx.) 0.7Vpp. That is an error of about 30%! In order to measure more correctly, use this rule of thumb: BW/3 equals about 5% error; BW/5 equals about 3% error. In other words: if the highest frequency you want to measure is 100 MHz, choose an oscilloscope of at least 300MHz, a better bet would be 500MHz. Unfortunately, this has the most influence on the price…
2 Understand that today’s signals are no longer pure sine waves, but most of the time square waves. These are built by “adding” the odd harmonics of the fundamental sine wave together. So a 10 MHz square wave is “built” by adding a 10MHz sine wave + a 30MHz sine wave + a 50MHz sine wave and so on. Rule of thumb: get a scope that has a bandwidth of at least the 9th harmonic. So if you’re going for square waves, it’s better to get a scope with a bandwidth of at least 10x the frequency of your square wave. For 100MHz square waves, get a 1GHz scope… and a bigger budget…
3 Consider rise (fall) time. Square waves have steep rise and fall times. There’s an easy rule of thumb to get to know what bandwidth your scope needs to be if these times are important to you. For oscilloscopes with bandwidths below 2.5GHz, calculate the steepest rise (fall) time it can measure as 0.35/BW. So an oscilloscope of 100MHz can measure rise times up to 3.5ns. For oscilloscopes above 2.5GHz up to about 8GHz, use 0.40/BW, and for scopes above 8GHz use 0.42/BW. Is your risetime the starting point? Use the inverse: if you need to measure rise times of 100ps, you’ll need a scope of at least 0.4/100ps = 4 GHz.
4 Choose your sample speed. Today’s oscilloscopes are almost all digital. The above steps involved the analog part of the instrument, before it gets to the A/D converters to get “digitized”. Here the bandwidth-to-rise time calculation can help you out: an oscilloscope of 500MHz has a calculated rise time of 700ps. To reconstruct this, you need at least 2 sample points on this edge, so at least a sample each 350ps, or 2.8Gsa/s (gigasamples per second). Scopes don’t come in this flavor, so choose a model with a faster sampling speed, e.g. 5Gsa/s (resulting in 200ps “time resolution”).
5 Decide on the number of channels. This is easy: most scopes come with 2ch or 4ch configurations, so you can choose what you need. Fortunately prices don’t double from 2ch to 4ch, but it does have a big impact on the price of the instrument. High-end scopes (>=1GHz) have always 4ch.
6 Calculate how much memory you’ll need. Depending on how much of your signal you want to see in a “single shot acquisition”, get your math right: at 5Gsa/s, you have a sample each 200ps. A scope with a memory of 10.000 sample points, can store 2µs of your signal. A scope with 100M samples (they do exist!) can store 20 seconds! Looking at repetitive signals or “eye-diagrams”, memory is less important.
7 Think about repetition rate. A digital oscilloscope uses a lot of time calculating. Between the moment of triggering (see next step), having the captured signal on the display, and capturing the next triggered event, most digital scopes “consume” several milliseconds. This results in only a few “photos” of your signal each second (waveforms per second), typically about 100-500. One vendor solved this problem with so called “Digital Phosphor” (from about 4.000 wfms/s to >400.000 wfms/s for the top models), others followed with similar-like technologies (but not always sustained/continuous, rather in bursts). This repetition rate is important because those rare errors and faults in your signal might occur just then when the scope is not acquiring, but busy calculating the last taken acquisition. The higher the repetition rate (wfms/s rate), the higher your chances are of capturing that rare event.
8 Check what kind of errors you expect to be looking for. All digital scopes have some sort of intelligent triggers on board, meaning you can trigger on more than just the rising or falling edge of your signal. If your repetition rate is high enough, you’ve probably seen that rare glitch every other second. Then it’s nice to have a Glitch trigger.
9 Think about resolution of LCD display. Small screens with poor resolution can make your life miserable if you cannot see results easily. Buy the largest screen with the best definition your budget will allow.
Some Final Tips
- Triggering, repetition rate and memory: once you found the rare event with a high wfms/s rate, having the right trigger available is more important than repetition rate, as your scope will trigger only on the (rare) event, which occurs… right: rarely. So you don’t need high rep-rate anymore. Memory can become more important, as to be able to analyze what happened before or after the event.
- Remember: garbage in is garbage out, so get the bandwidth and rise time issue sorted out first!
Established in 2004, NorthTree Associates (Cologne, MN) is a North American distributor that specializes in providing unique Electronic Test & Measurement tools for design engineers, test engineers and production engineers. You can visit our website at http://www.northtreeassociates.com